Enactivism · Predictive Coding · Active Inference · Affordance Perception
2018
Formalising Affordances: Dopamine-Regulated Cue Salience
A dopamine-regulated cue salience model of selective affordance perception in Bipolar Mania, extending two existing accounts to a case neither had addressed.
Client
Completed · (student) conference presentation
scope
Enactivism · Predictive Coding · Active Inference · Affordance Perception
year
2018
duration
Overview
The question.
Enactive cognition holds that a mind is not a passive receiver of the world but something that actively creates, or enacts, a world through its ongoing interaction with the affordances an environment offers. A long-standing objection to this view is that it remains largely descriptive — compelling as phenomenology, but difficult to test. Friston et al. (2012) had shown that dopamine-regulated cue salience could give a computationally and neurobiologically plausible account of affordance perception in Parkinson's disease, and Bruineberg and Rietveld (2014) had used a related framework to characterise the distorted affordance perception found in OCD. This paper asks whether the same mechanism — dopamine's role in weighting the precision of action-soliciting cues — can be extended to a disorder neither account had addressed: the manic phase of Bipolar disorder.
The Challenge
Two accounts, two disorders, one convergence.
Friston's model and Bruineberg and Rietveld's account offer different levels of explanation — one computational and biologically grounded, the other phenomenological and qualitative — and had, until this point, been applied to different conditions. My argument is that Bipolar Mania sits at their intersection: patients in the manic phase both overestimate the time delay on rewards, making immediate cues more salient, and show evidence of overly precise priors, a form of aberrant predictive coding that makes previously-rewarded cues more salient too. Where these two forms of salience converge, I argue, is where selective openness to affordances in BPM is best explained.
the solution
Computational enactivism, tested.
Rather than treat enactivism's abstractness as a weakness to explain away, I propose two concrete, falsifiable hypotheses generated by the model: that midbrain dopamine projections should be more pronounced when an affordance offers both an immediate and a previously-experienced reward than when it offers only one, and that individuals in a hypomanic state should show less exploratory behaviour, favouring familiar rewards over equally valuable unfamiliar ones. Both are testable with fMRI, using an experimental design comparable to Schwartenbeck et al. (2014).
